How can you generate and validate keywords from customer language?

To generate keywords from customer language, start by organizing pain points and direct quotes from reviews, discussion threads, and sales calls. Turn these findings into potential search queries, then use search volume, keyword difficulty, and search intent to help prioritize them. What customers say offers ideas for content, but does not confirm search demand; a tool reporting zero search volume does not automatically mean there is no demand. Keep a source excerpt for each candidate keyword so you can check whether it preserves the customer’s original meaning.

Start by gathering context around problems, not just product names

The value of this material lies in how customers describe problems, what they complain about, and which features they praise or want. Choose sources based on the type of product:

• B2B software: Look at competitor reviews on platforms such as G2 and Capterra.
• B2C products: Look at reviews on Amazon, competitor websites, and other retail websites.
• Sales calls and Reddit: Organize pain points, usage limitations, and feature requests, keeping the relevant conversations or comments.

Paid search queries also provide authentic wording, but short phrases offer limited context. Call transcripts provide more context, but should not all be fed into a model at once. Seer recommends first tagging signals such as objections, competitor mentions, and decision criteria, then retrieving direct quotes as needed.

Group by topic, then turn the findings into candidate queries

First, identify recurring themes in the reviews, then extract direct quotes that support each theme. Finally, think of phrases that someone facing those problems might search for. Your tracking table can include candidate keyword, theme, source, and original excerpt; do not keep only an AI-generated keyword list.

At Procrustes, our content services turn the questions customers genuinely care about into articles, with ongoing planning, writing, and updates.

For example, if a competitor review mentions a missing feature that your product genuinely offers, you could consider content explaining that feature. However, this is only a content opportunity, not a guarantee of rankings or AI citations. Customer wording found through organic search can also inform paid ad copy, so you do not have to force customers to use internal product terminology.

Validate the data and check search intent

Once you have gathered the candidate keywords, you can use Semrush to check search volume, keyword difficulty, and intent to help set priorities. Similar wording does not necessarily mean the same search intent: searches for “what is search engine marketing” and “search engine marketing agency” call for different content and have different commercial value.

Keywords that Reddit threads already rank for can offer clues, but this method depends on those threads ranking well on Google and may miss other ideas. You should therefore still read the full discussions and comments rather than collecting only ranking keywords.

When search volume is zero, revisit the original wording and check variations

If Semrush shows zero search volume or no data, it may simply mean that no search volume is currently shown for that exact wording. Keep this limitation in mind. Check whether the problem is relevant to your business, then develop variations and check again. When the tool provides no intent data, any intent inferred from context should also be labeled as an estimate.

If a customer question is worth answering, you can still consider expanding existing content or writing a separate article, even with limited data. But do not present “worth answering” as “proven to have high search volume.”